Roundtable 10-26 on Dictators and Their Secret Police: Coercive Institutions and State Violence83 min read

With apologies to Tolstoy, every coercive dictatorship is coercive in its own way. This is the central claim of Sheena Greitens important and timely study of authoritarianism, Dictators and Their Secret Police. Greitens argues that dictators face not only the usual array of external threats that all leaders confront; they also face a daunting array of violent internal threats that can range from elite-led coups all the way to general popular uprisings. Different dictators perceive and prioritize these threats differently, and adjust the design of their institutions of state coercion accordingly. These different configurations of state coercion yield their own differential implications for the political choices confronting citizens and their likely behavior in response. Ultimately, Greitens argues that she can trace political behavior all the way from a leader’s perceptions of threat to the response of citizens, thus providing a theory of institutional design and evolution. She demonstrates this claim with detailed studies of the evolution of dictatorships in the Philippines, Taiwan, and South Korea, and more summary considerations of Iraq, East Germany, and Chile.

H-Diplo | ISSF Roundtable, Volume X, No. 26 (2019)

Editors: Caitlin Talmadge and Diane Labrosse
Web/Production Editor: George Fujii

Sheena Chestnut Greitens. Dictators and Their Secret Police: Coercive Institutions and State Violence. Cambridge: Cambridge University Press, 2016. ISBN: 9781107139848 (hardcover, $105.00); 9781316505311 (paperback, $30.99).

Published on 15 April 2019



© Copyright 2019 The Authors


Introduction by Peter D. Feaver, Duke University

With apologies to Tolstoy, every coercive dictatorship is coercive in its own way. This is the central claim of Sheena Greitens important and timely study of authoritarianism, Dictators and Their Secret Police. Greitens argues that dictators face not only the usual array of external threats that all leaders confront; they also face a daunting array of violent internal threats that can range from elite-led coups all the way to general popular uprisings. Different dictators perceive and prioritize these threats differently, and adjust the design of their institutions of state coercion accordingly. These different configurations of state coercion yield their own differential implications for the political choices confronting citizens and their likely behavior in response. Ultimately, Greitens argues that she can trace political behavior all the way from a leader’s perceptions of threat to the response of citizens, thus providing a theory of institutional design and evolution. She demonstrates this claim with detailed studies of the evolution of dictatorships in the Philippines, Taiwan, and South Korea, and more summary considerations of Iraq, East Germany, and Chile.

In this roundtable, four scholars of authoritarianism and civil-military relations offer their careful evaluations of Greitens’s arguments and evidence. They praise Greitens for the elegance and parsimony of her theory. They note the care with which she has conducted her empirics, albeit whilst offering differing interpretations of this or that specific incident. They identify this book as an important contribution to the burgeoning literature on the politics of authoritarian regimes and commend it to scholars across a range of subfields. At the same time, they raise questions on both the book’s theory and the empirics, and identify some important next steps that make clear they do not consider this work to be the last word on the subject.

Risa Brooks, a specialist in comparative security politics, offers a balanced but strongly positive evaluation of the book. She credits Greitens with making a “major advance” in how scholars theorize state repression and focusing attention on the phenomenon of when state repression is so effective that it manifests in a coercive peace. Brooks similarly suggests that Greitens’s work on the origins of mass protest helpfully theorizes both the occurrence and non-occurrence of protests that can lead to widespread violence. In short, Brooks finds this book to be “compelling and significant,” though she also finds that it leaves unanswered other, equally compelling, questions. Brooks doubts that dictators face as limited and binary a set of choices as Greitens’s model posits. Likewise, drawing on her own work, Brooks questions whether it makes sense to treat military and non-military security services as equivalent. She also raises the issue of the agency of these security actors, a major focus of Joseph Wright’s review (see below).

Jasen Castillo, a specialist in comparative military institutions, offers a particularly favorable assessment of what Greitens has done, while identifying important research questions for what Greitens has not done. He considers this book to be the new baseline in the study of authoritarianism, one that scholars will ignore “at their peril.” Yet, he also sketches a research program to answer questions focused around variables and dynamics that Greitens does not examine: how ethnic cleavages within societies complicate the dictator’s institutional design challenge; the conditions under which the leaders of popular uprisings necessarily collude with potential coup plotters and the effects such potential collusion might have on the strategic calculations of dictators; and how external threats might interact with the dynamics Greitens outlines.

Van Jackson, a specialist in northeast Asian security (among other topics), is effusive in praising Greitens work, both for its internal contributions and for how he sees it applying to a case Greitens does not consider very closely, but that is very much in the news: North Korea. He applies Greitens’s model to North Korea and suggests it implies that the current Kim regime may be more vulnerable to a coup and less vulnerable to popular uprising than many outside observers believe. He notes that Greitens’s analytical lens could lead to the opposite inference, depending on how experts assess the efficacy of Kim Jong-un’s efforts to kill off potential rivals. Either way, Jackson believes Greitens’s approach yields insights that would benefit Korea watchers. That said, Jackson notes that Greitens bracketed off from her theory an important set of tools that any dictator would have to consider, namely the carrots of cooptation to go along with the sticks of coercion. He speculates that adding in these other tools would enhance the predictive power of the model to assess when leader-society relations will turn violent.

Joseph Wright, who has made his own important contributions to the study of the domestic politics of authoritarian regimes, offers a mixed assessment of Greitens’s book. He praises her for shifting the analytical lens earlier along the causal pathway, looking not only at how coercive institutions shape political behavior but also at the political dynamics that give rise to the institutions in the first place. He also credits Greitens for writing well-crafted case studies that make effective use of archival evidence. At the same time, however, he critiques Greitens theory as decision-theoretic, when a game-theoretic approach is preferable. Specifically, he claims that the security institutions (armies, police, intelligence forces) themselves anticipate and react to efforts by dictators to constrain them, and approvingly cites two works by Jun Koga Sudduth approvingly to argue this point. Wright further suggests that some of Greitens’s own evidence may better fit the opposite predictions of the game-theoretic model than of her decision-theoretic approach. On the empirics, he notes that almost all of her cases involve dictatorships that evolved peacefully into democracies. Since 1945, however, most dictatorships have either retained power intact, given way to other dictatorships, or ended with violent regime change. He suggests, therefore, that the question of how generalizable Greitens’s argument is must be left pending until further work is done on cases drawn from these other categories.

Collectively, these reviews and the underlying work point to a conclusion that should encourage both satisfaction and humility: our understanding of important political phenomena involves discoveries that increase our knowledge and simultaneously increase our awareness of what we still do not know—often in equal measure. Sheena Greitens has made an important contribution, by shining a powerful analytical light on dynamics that are only dimly understood: the political and institutional choices of autocrats. In the process, she has pointed the way to future work that will extend, refine, and, perhaps in certain respects, one day overtake even this significant piece of scholarship.


Sheena Chestnut Greitens is an assistant professor at the University of Missouri and a non-resident fellow at the Brookings Institution and the Korea Chair at the Center for Strategic and International Studies. Her work focuses on security, authoritarian politics, and East Asia. Dr. Greitens holds a Ph.D. from Harvard University; an M.Phil. from Oxford University, where she studied as a Marshall Scholar; and a B.A. from Stanford University. Dictators and Their Secret Police, her first book, received the 2017 best book award from the International Studies Association as well as the Comparative Democratization section of the American Political Science Association. She is currently working on projects on contemporary China and North Korea.

Peter Feaver is a Professor of Political Science and Public Policy at Duke University, where he directs the Duke Program in American Grand Strategy and the Triangle Institute for Security Studies

Risa Brooks is Allis Chalmers Associate Professor of Political Science at Marquette University where she specializes in the study of civil-military relations (comparative & American), military effectiveness, political violence and domestic terrorism in the United States. She is also presently an adjunct scholar at West Point’s Modern War Institute (2016-2017; reappointed for 2017-2018). Professor Brooks is the author of Shaping Strategy: the Civil-Military Politics of Strategic Assessment (Princeton University Press, 2008) and editor (with Elizabeth Stanley) of Creating Military Power: the Sources of Military Effectiveness (Stanford University Press, 2007). Her current book project (with Peter White) explores the causes and consequences of the strategies through which dictators maintain political control of the military in authoritarian regimes.

Jasen J. Castillo is an associate professor and Evelyn and Ed F. Kruse ‘49 Faculty Fellow in the Bush School of Government and Public Service at Texas A&M University. Prior to joining the Bush School, he worked in the Department of Defense’s Policy Planning Office and at the RAND Corporation. He is the author of Endurance and War: The National Sources of Military Cohesion (Palo Alto: Stanford University Press, 2014).

Van Jackson is a Senior Lecturer in International Relations at Victoria University of Wellington, Defence & Strategy Fellow at the Centre for Strategic Studies, and Global Fellow at the Woodrow Wilson International Center for Scholars in Washington. He is the author of two Cambridge University Press books on U.S.-North Korea relations, the most recent of which is On the Brink: Trump, Kim, and the Threat of Nuclear War (2018). He is also the writer and director of a forthcoming documentary called The Nuclear Button: How Trump and Kim Blustered to the Brink of War.

Joseph Wright is political scientist at Pennsylvania State University. He studies comparative political economy with a particular interest in how international factors—such as foreign aid, economic sanctions, human rights prosecutions, and migration—influence domestic politics in autocratic regimes. His first book, with Abel Escriba-Folch, Foreign Pressure and the Politics of Autocratic Survival (Oxford University Press, 2015), won the 2017 Stein Rokkan Prize for Comparative Social Science Research. His second book, with Barbara Geddes and Erica Frantz, is How Dictatorships Work (Cambridge University Press, 2018).


Review by Risa Brooks, Marquette University

Sheena Chestnut Greitens has written an important book that is bound to have a major impact on the fields of comparative politics and security studies. The book focuses on the implications of what Greitens calls the autocrat’s “coercive dilemma”: leaders must choose whether to create coercive institutions that prevent elite (coup) threats or those that limit popular opposition.[1] The choice hinges on an autocrat’s “perceived threats.” Where autocrats perceive coup threats as dominant over those posed by popular unrest, they create fragmented and socially exclusive institutions (i.e., those that are competitive and unrepresentative of society). Where fears of popular opposition prevail, they create unitary and inclusive coercive institutions (those with clear jurisdictions, central authorities, and that are representative of society). As such, Greitens argues that autocrats face a stark choice in coercive institutional design: a fragmented and exclusive set of institutions cannot undertake efficient repression, and unitary and inclusive institutions cannot prevent coups (see, for example, 32).

These choices, moreover, are not only consequential for the autocrat, but for the society he or she rules. In the second part of the book, Greitens shows how the structure of coercive institutions shapes patterns of repression and violence against civilians. Fragmented institutions are apt to result in repression that requires high levels of violence, whereas unitary systems can use repression more surgically and efficiently. In this way, the institutional choices autocrats make about their security forces have life or death implications for the dictatorship’s citizens.

The book’s theory is well conceptualized and the empirical analysis is smartly conceived and executed. Greitens’s analytical chapters elucidate the causal logics underpinning elements of the theory. She then creatively and judiciously uses evidence in support of different steps along the causal chain. The empirical chapters consequently link seamlessly to the theory. For that reason alone, the book should be read by graduate students and for those of us working on major research projects—it represents a model of how to execute process tracing and qualitative analysis. In short, the book is a terrific read and an accomplished piece of social science.

The book makes several contributions to academic debates, which I discuss below. I then turn to several questions and concerns raised by the analysis.

Among the book’s central contributions is its focus on variation in state-led violence against civilians. There is a sizable literature on civilian violence by non-state actors in insurgency and terrorism. As Greitens rightly notes, however, considerably less systematic research exists on state instigated repression (6-11). In particular, the book advances our understanding of how to conceptualize and measure the dependent variable of ‘state repression.’ Greitens defines variation in state violence according to its scope, intensity and (in)discriminate nature. She then operationalizes those concepts, looking to measures such as regional patterns of violence, the number of people killed and arrested, the collective versus individual nature of arrests, and the use of extrajudicial killings (65). The book thus represents a major advance in the study of political violence in its variations incarnations.

Second, Greitens provides insight into what might be termed the ‘effectiveness’ of policing and repression in autocratic regimes. Scholars have long argued that civil-military relations and coup-proofing can impair military effectiveness in war against foreign adversaries (41). Greitens’s argument shows how civil-military relations and coup-proofing can similarly undermine effectiveness in internal policing. As she demonstrates, some coercive apparatuses efficiently gather intelligence and apply repression selectively in ways that anticipate protest and opposition, and therefore deter and prevent it. Others are less capable and resort to overt violence to secure autocratic rule. Empirically, this underscores how differently the mechanisms and form of repression can manifest in autocracies. Normatively, it reminds us that the absence of overt violence does not equate to an absence of repression (and therefore more legitimate governance); rather, repression is undertaken with greater effectiveness by regimes whose autocrats have generated institutions for that purpose.

A third significant contribution is the book’s potential to explain the origins of mass protest. It addresses what might be understood as ‘the endogeneity of mass protest to the nature of coercive institutions,’ that is, whether or not mass protests occur might depend on how effectively repression is carried out or on other features of intelligence and security institutions. There is already some evidence of this dynamic in the Arab Spring protests of 2011; in particular, there are hints of it in Syria, which has fragmented and socially exclusive coercive institutions. In March 2011, several teenagers in Dara’a were subjected to brutal violence by security forces after their relatively mild anti-regime graffiti was discovered. Their harsh treatment, in turn, helped catalyze a mass uprising, which at that point no one foresaw as occurring in Syria (including the initial protesters).[2] In the final pages of the book, Greitens highlights these implications; she notes that if indiscriminate violence causes opposition, then understanding the origins of the coercive institutions that cause that violence might illuminate why and when protests occur (305). Her argument may therefore help explain both why protests do occur, and why they do not (i.e., the non-events).

These important contributions aside, some of the analytical choices Greitens makes in the book merit consideration. I discuss three of those choices below. Before proceeding, however, one clarification is in order: my intention in posing the questions below is not to challenge Greitens’s core argument or the book’s findings. To the contrary, they are compelling and significant. The aim instead is to highlight the choices she has made in her research and to raise questions about the implications of those decisions. Social science research requires scholars to decide how to conceptualize variables and their relationships, make causal claims, and articulate scope conditions. Greitens’s lucid narrative and methodical approach to theory development makes these decisions very evident to the reader. In so doing, she has provided us much food for thought.

To start, Greitens makes a rather strong claim about the institutional options that are available to leaders who try to prevent coups (24-25, 32). She argues that when faced with a dominant coup threat, autocrats are forced to fragment and engage in social exclusion in their coercive institutions. They must create in-groups and competition to prevent coups. While she allows that this might not be true in every case, and that there are other ancillary methods autocrats use, she concludes that these are the go-to methods of coup-proofing.

But is the choice that autocrats face as stark as she claims? While fragmentation and social exclusion may be common solutions to coup threats, there are other strategies of political control upon which autocrats can and do rely—that is, other means by which leaders prevent coups, while retaining executive power over the state. In Egypt, which Greitens references, leaders did not fragment the coercive apparatus. Rather, President Hosni Mubarak quite ably prevented a coup for three decades by forging what might be termed a “grand bargain” with the military.[3] Similarly, in some formerly Communist countries in Eastern Europe, coups were prevented with a rather centralized or monolithic coercive apparatus, with clear lines of command and divisions of responsibilities. Extensive monitoring by party elites and the professionalization of the military were essential in those cases.

Greitens might counter that these autocrats did not face high perceived threats of elite coups, so they could afford to employ these alternative coup-prevention strategies, rather than create fragmented and exclusive coercive institutions. But that begs the question of why these autocrats did not perceive high coup threats. If they had developed a strategy for political control that works, we would not expect them to perceive high coup threats (barring some exogenous change that generated the threat). This points to a limitation in the inductive approach to measuring “perceived threat” upon which Greitens relies. It may obscure the endogeneity of coup threats to unitary coercive institutions.[4] In other words, if some autocrats have figured out ways to mitigate coups within unitary systems (or other institutional variants), researchers may as a result code them as cases of low perceived coup threat. This could create some circularity in that Greitens would predict these autocrats would not prioritize coups and therefore would adopt unitary coercive institutions. In short, the two coup-proofing mechanisms on which she focuses—fragmentation and social exclusivity—may not be as indispensable to autocrats as the book contends.

A second curious analytical choice relates to the decision to lump the military and non-military intelligence and security services together and treat them as uniform types of actors. This might strike many scholars of civil-military relations oddly. It elides important differences in the functional roles, organizational interests, and sociology of militaries versus police.[5] Police and security forces and conventional militaries in autocracies often have rivalries and divergent interests that are distinct from intra-police/security force fights, which could affect the development of the coercive institutional sector.[6] Political leaders may employ different strategies of control toward their militaries versus their police and security sectors.[7] Police and military also might have different decision-making criteria and roles in repression. In the Arab uprisings of 2011, police across the board engaged in repression when protests occurred, but the response by militaries varied across the uprisings.[8]

This relates to another critical decision—treating institutional outcomes as the pure projection of the autocrat’s interests. The military or police have little or no agency or power in the story Greitens tells. Yet, we know from other research that institutions in the coercive sector can be shaped by negotiation and power relations between military and political leaders.[9] Given that autocrats often rely on militaries as key players in their ruling coalitions, this one-sided story is puzzling. The possibility that coercive institutions depend on a bargaining relationship between military/security force chiefs and the autocrat might be considered an alternative explanation. Alternatively, perhaps a scope condition might be attached to the analysis such it applies to cases where the autocrat has dominant political power to dictate institutional outcomes.

Finally, the decision to treat the two parts of the analysis as essentially separate arguments raises some questions. For example, how do the outcomes studied in part two of the book (violence against civilians) bear on the decisions autocrats make about institutional design that are discussed in part one? The autocrats in Greitens’s narrative are pretty savvy operators. If indeed it is the case that fragmented institutions and violence against civilians could generate the threat of popular unrest, or at least further undermine the legitimacy of authoritarian rule, autocrats might foresee this. Do they understand the risks they are taking? Why don’t they try to manage them better? Rather than optimize institutional design to address one threat, as Greitens contends, might they instead satisfice to try and balance the risks inherent in different institutional choices? In fact, these tradeoffs might be one reason that autocrats innovate different means of coup-proofing other than fragmentation and exclusivity.

These comments and questions aside, there should be no doubt that this is an extremely thoughtful and carefully argued book. Greitens has made a major contribution to our understanding of political violence, civil-military relations, and autocracy.


Review by Jasen J. Castillo, George H.W. Bush School of Government, Texas A&M University

Sheena Chestnut Greitens’s impressive book provides an answer to an important puzzle in the field of comparative politics: How do authoritarian leaders create and then use coercive institutions to preserve their rule? Her study outlines the strategic logic facing dictators as they determine their best strategy to counter domestic threats. Dictators must navigate many trade-offs when deciding the ideal approach to securing their positions. Using a simple and elegant theory, Greitens explains that every autocrat must decide whether to focus on either the danger of a popular uprising or an elite coup d’état, in which the military topples the regime.

To prevent a mass revolt, dictators will, according to her theory, create unitary and inclusive coercive institutions. These organizations result in low levels of violence, as government agents try to win the hearts and minds of the general population. Here, I am reminded of the former German Democratic Republic. The logic of this approach explicitly embraces the tactics of counterinsurgency. In contrast, when they fear elite coups, dictators fashion fragmentary and exclusive repressive organizations. Recall the divide-and-conquer tactics of President Saddam Hussein’s regime in Iraq. These institutions foster higher levels of violence, as government security services often work at cross-purposes and terrorize excluded segments of the population. The logic of this approach reflects recent work on the coup proofing of authoritarian states.[10]

Dictators and Their Secret Police will demand the attention of anyone interested in the depressing but serious subject of authoritarianism. Not only is the book written carefully and clearly, but readers will also find at least two significant contributions to the study of dictatorships. First, Greitens deduces a straightforward theory that simplifies the strategic decision-making of dictators. She isolates the key threats authoritarian governments confront, bracketing less important concerns. I especially appreciated the transparent causal logic connecting perceived threats to the choice of coercive institutions and the resulting violence. Second, Greitens demonstrates the plausibility of her theory through well-crafted and considered case studies of dictatorships from East Asia. Readers do not have to slog through potted history. Nor are these the drive-by historical narratives that seem to plague much of political science today. Instead, the historical work, reinforced by archival research, thoughtfully shows the power of the book’s theory, while retaining the good sense to admit when the argument does not fit the facts perfectly. For these reasons, practitioners of U.S. foreign policy trying to discern the intentions and understand the behavior of dictatorships will also find much to like in this book.

Because Dictators and Their Secret Police presents such a clear argument, it is easy to engage with the theory. As the field of comparative politics digests these findings, scholars will begin to ask new questions that are either motivated by the book’s conclusions or are left unanswered by Greitens’s study. My questions focus on variables omitted from the theory. None of these questions, however, diminishes the book’s contribution. Instead, they suggest future research that builds on the excellent foundation laid down by Greitens.

First, how and under what conditions would ethnic divisions alter a dictator’s calculations? To streamline a dictator’s strategic decision-making for countering domestic threats, the book’s theory sets aside foreign dangers and secessionist movements. Understandably, this research design narrows the scope of the theory and, therefore, makes the project manageable. Nevertheless, I am curious how ethnic divisions that fall short of secessionist movements might change a dictator’s choice of strategy. These types of divisions have plagued authoritarian leaders of post-colonial regimes.[11]

Second, what are the ways in which a popular uprising would end a dictatorship? A few historical examples in the book’s theory chapter would help readers understand how a large-scale revolt would upend an authoritarian regime. The examples that come readily to mind include the collapse of regimes in Eastern Europe at the end of the Cold War, especially Romania and East Germany. I raise this issue about the collapse mechanism during a popular revolt because it seems plausible that the success of such an uprising requires at least the tacit consent if not explicit help of the armed forces. If true, this observation suggests that the threat posed by a popular uprising and a military coup might share more characteristics than implied by the book’s theory.

Third, and related to the point above, why does the theory exclude the possibility that a dictator could worry simultaneously about both a popular uprising and a military coup? If the book’s argument did include this option, then I suspect authoritarian leaders would closely follow the coercive technique for thwarting an insurrection by the armed forces. After all, securing the obedience of the armed forces seems a necessary condition for regime survival. Coup proofing, then, would represent the central tenet of any strategy that sought to counter both elite threats and a popular revolt. The problem, however, is that Greitens’s theory defines coup proofing as a strategy that is incompatible with preventing popular uprising. While some dictators may face such a stark choice, I can imagine instances where authoritarian leaders needed to address both threats.

Consider, for instance, the choices made by the dictators of Nazi Germany and the Soviet Union. Of course, I can appreciate Greitens’s desire for moving beyond these two long-studied dictatorships, but her theory should shed light on not only on new, less explored cases but on the familiar ones as well. Adolf Hitler and Joseph Stalin employed coup-proofing schemes to protect their respective regimes. In the Soviet case, this strategy undercut the Red Army’s military effectiveness at the start of the Second World War.[12] Additionally, both regimes worried about the potential for popular uprisings to undermine their regimes. Now, there were important differences in how they envisioned this danger, but the key point is that the two regimes feared a popular revolt enough to take actions to prevent it.[13]

In particular, they relied on a combination of terror and compelling nationalist ideologies to guard against popular uprisings. By destroying civil society, both regimes ensured that their respective ideologies took root and found hard-core supporters to enforce it. In this way, powerful ideologies married to coercion permitted both regimes to exercise great control over their societies. [14]  This history suggests that not only have dictators acted to counter both coups and popular revolts simultaneously, but they also have a strategy for doing so.

Finally, how do external threats influence repression strategies? Does the presence of a dangerous foreign enemy decrease incentives for elite or mass revolts? In a few of the book’s historical cases, we see evidence of that international pressure influenced a dictator’s decision-making. Future researchers might want to push Greitens’s theory by explicitly incorporating external threats to see if these worries interact with how authoritarian regimes approach domestic dangers.

None of these questions detracts from this significant mark that Dictators and Their Secret Police will leave on the field of comparative politics. On the contrary, Greitens has created the new baseline that the rest of us will need to grapple with when we study authoritarianism. The book’s likely influence on the field is a testament to the clarity of her theory, research design, and historical case studies. When we want to understand the strategic behavior of dictatorships, we should start here. Scholars will ignore this book at their peril.


Review by Van Jackson, Victoria University of Wellington and The Centre for Strategic Studies

It’s hard to hear the word dictatorship without immediately imagining fear, violence, and repression. Yet, as Sheena Greitens points out, the bulk of research on the inner workings of authoritarianism tends to overlook “the variation that exists across the institutions that manage domestic surveillance, repression, and violence” (292). Her comparative examination of “coercive institutions”—internal security, police, domestic intelligence services, and elements of the military—addresses this gap in an otherwise wide-ranging, highly saturated literature on autocracies.

In brief, Dictators and Their Secret Police puts forward two arguments. First, the design of coercive institutions is a function of a dictator’s dominant perceived threat. When dictators believe the greatest threat emanates from elites (especially security elites), they optimize their coercive institutions against coups by crafting them in deliberately fragmented, socially exclusive ways; when the masses pose the greatest threat, dictators forge unitary and cohesive coercive institutions. Second, it argues the pattern of violence within dictatorships systematically reflects the design of coercive institutions. Fragmented and competitive coercive institutions lead to higher levels of violence, while cohesive ones use violence more discriminately.

This is qualitative, positivist research at its best. The argument is intuitive, parsimonious, and empirically substantiated across a wide array of cases. It occupies a breach in the comparative literature on autocracies. And as I discuss below, it’s the kind of middle-range theorizing that has real-world tractability for those casting about for a solution to North Korea, perhaps the world’s most intractable policy problem. Greitens has given us a relatively simple way of explaining an empirical regularity—the design of coercive institutions and the effects of their repressive tactics—better than competing ‘most likely’ alternative theories.

The biggest questions that the book raise have less to do with ‘what’s in it’ (which is quite good) than with what it omits. These questions do not represent criticisms so much as opportunities to either refine or build on her findings.

The Coercive vs. Consensual Dilemma

While the book interrogates patterns found within the spectrum of coercive institutions, and their corresponding behavior, those boundaries raise an even more interesting puzzle whose answer could have superior explanatory power: How and why do dictators decide among the various carrots and sticks at their disposal to keep themselves in power? Dictators not only face a “coercive dilemma” when it comes to optimizing their internal security organizations; they also face a dilemma in selecting among consensual and coercive approaches to their plight (preserving their regime control) generally. No modern autocratic regime stays in power on fear alone, and in principle they can manipulate affect, greed, or anger rather than fear. Dictators rely on any number of non-coercive tools—national myths, rituals, information controls, and material factors like outright cronyism—in addition to, or possibly even as substitutes for, optimizing coercive institutions in a manner that blunts the potential for coups or popular revolts. These alternative ‘consensual’ tools of control preemptively narrow the range of legitimate challengers to power, and can have significant sway over the popular desire to challenge central authority. But how do dictators mix and match?

Admittedly, this larger question on the ‘carrots’ and ‘sticks’ of regime security is outside of what the book tries to explain, but it matters because the answer may give dictators a way around—or at least a way of managing—the coercive dilemma they face at the organizational level. The book offers a brief discussion about why it chooses to avoid this more expansive line of inquiry: “Autocrats are unlikely to see co-optation and cronyism as sufficient insurance against coup risks…” (24). But the existence of a threat often does not determine the response to the threat. Greitens reasons that even if a dictator pursues multiple means of regime security it still has to decide how to design its internal security establishment. True enough, but the ability of a dictator to draw on non-coercive tools of domestic statecraft can distort patterns of domestic unrest. It may well be that ‘carrots’ (or non-coercive variables) facilitate how dictators resolve the important ‘stick’-based coercive dilemma that Greitens presents. And even if Greitens gets the answer to her first question about patterns of coercive institutional design right (and I think she more or less does), the ability to explain the degree and kind of domestic violence that result (her second question) could still be limited only to those cases where a dictator has failed to address threats through measures other than the internal security apparatus.

Reputation Effects?

A different question also emerges from Greitens’s second argument (about patterns of violence): To what extent are patterns of internal violence endogenous to the historical use of internal violence? Might there be a hidden reputational mechanism that enhances deterrence against popular dissent at play within her theoretical argument? The book’s secondary argument basically predicts that violence will become rarer and more selective in regimes whose coercive institutions are socially inclusive and cohesive, because 1) they are more effective in collecting and analyzing intelligence against regime dissenters, and because 2) they have fewer incentives to engage in competitive violence with other institutions. But it seems equally plausible that the efficacy of unified coercive institutions has a deterrent effect on the population that then reduces the need for violence, which would suggest that an added reputational mechanism intervenes somewhere between the independent variable (character of the institution) and dependent variable (pattern-of-violence outcome) over time. This does not undermine Greitens’s empirical observations, but highlights a possible alternative mechanism within her theory that explains the broader correlation between institutional configuration and its uses of violence.

Exploiting the Coercive Dilemma in Practice

Beyond its scholarly merits, the book’s theoretical framework also lends itself to reorienting analytical questions about North Korea and how best to deal in practice with the regime governing it. In the concluding chapter, Greitens seems to downplay the explanatory potential of her argument for North Korea, but the “coercive dilemma” frame opens up interesting possibilities for better analysis of—and, by extension, ways of thinking about policy toward—the North that are orthogonal to the traditional menu of sanctions, deterrence, and diplomatic negotiations.

Kim Jong-un inherited a highly fragmented and socially exclusive security regime from his father, Kim Jong-il, but he appears to be in the process of unifying it. The elder Kim was overwhelmingly obsessed with coup-proofing, not least because he faced threats from the military and security services (there were reportedly multiple coup attempts in the 1990s).[15] But since taking power in 2011, it is not clear that Kim Jong-un sees institutional coup-proofing as altogether necessary. To the contrary, Kim Jong-un has overseen more than 300 executions in the past six years,[16] mostly of senior ranking officials from throughout the regime (including his once-powerful uncle, Jang Song-thaek). Kim Jong-un has systematically replaced purged officials with loyalists, many of whom shared direct personal ties to Kim rather than only to his father or grandfather. Most Korea watchers interpret these killings as an attempt to consolidate control over his security institutions, rather than deliberately fragmenting them as his father had done. This is useful information in at least three ways.

First, if we accept that Kim Jong-un is moving away from the fragmented system of coercive institutions of his father, then in relative terms his regime is becoming less vulnerable to popular disruption. This raises doubts about the effectiveness of the ‘asymmetric’ approach to North Korea whereby the United States seeks to penetrate the country’s information controls by smuggling in and proliferating knowledge and entertainment from the outside world. This approach to North Korea became so attractive during the Obama administration that it became a long-term project in the State Department,[17] and Silicon Valley “hackathons” sponsored by human rights groups looked for technological solutions to doing so.[18] Yet even if attempts to subvert the regime by spreading information succeeds in creating social uprisings, the result may simply be a ruthless bloody crackdown by an increasingly cohesive and efficient security apparatus. The implications for how the coercive dilemma framework applies to the North Korea situation thus suggests tempering enthusiasm for what many see as a “cost-free” response to the North Korea challenge; we should be wary, or at least witting, of the unintended consequences of encouraging a revolt in a country with highly effective, repressive security institutions.

Second, again proceeding from the premise that Kim Jong Un is unifying his coercive institutions, Kim is, in relative terms, more vulnerable to ouster by coup than he would otherwise be. It is doubtful that outside powers could facilitate regime change, and it’s not obvious that doing so would even be desirable, but if it is incrementally more likely to occur, then it should be added to the analytical obsessions of ‘Korea watchers’ in the United States and South Korea—even if neither wishes to agitate for a coup process. Interested constituencies within government and outside it should be working through the many permutations of who might be involved in coup scenarios and how it might affect North Korean foreign and defense policy; as far as I’m aware, this has not been an analytical priority for Korea experts anywhere.

Finally, it is entirely possible to draw the opposite interpretation of the above—that Kim Jong-un’s wave of assassinations is an attempt to create a fragmented, not centralized, security apparatus. It could be argued that Kim Jong-un is just like his father—consumed by the need to coup-proof his regime. The assassination wave could be proof of paranoia and insecurity, and his insertion of loyalists as substitutes for the purged could be setting up deliberately competitive camps. But even this contrarian interpretation of North Korea illustrates the larger value of the coercive dilemma framework, as an important axis of analytical contention among Korea watchers. Very few of the debates within the community of Korea experts link competing interpretations of evidence to well-reasoned theoretical frameworks (or even to poorly reasoned ones). The coercive dilemma can and should serve as a new axis for debate about regime stability and opportunities for influence.

Non-traditional approaches to North Korea are possible. We could pursue a ‘divide-and-rule’ strategy, engaging directly with competing elite power centers in circumvention of North Korea’s Foreign Ministry. We could, alternatively, pursue a ‘subversive engagement’ strategy, where we somehow disseminate grievance-inducing information to the masses. In either case though, the preferred approach depends in part on interpretations of how the regime is structured—especially its coercive institutions. This goes beyond the book’s original intent, but Dictators and Their Secret Police is a sorely needed baseline for sharper thinking about one of the hardest policy problems facing the United States today.



Review by Joseph Wright, Pennsylvania State University


How do dictators design their coercive apparatus? And how does this design influence the level and pattern of state-led violence? Sheena Chestnut Greitens’s fascinating book takes up these questions using an array of evidence from four East Asian regimes: Ferdinand Marcos’ regime in the Philippines; the Kuomintang (KMT) regime in Taiwan; and two periods of military rule (Park Chung-hee and Chun Doo-hwan) in South Korea. The book starts with the central threats all dictators face: those from outside the regime and threats from within the security forces. These threats, Greitens argues, determine how dictators design their security forces, with implications for state-led violence and repression. Building a theory from these two threats puts this book firmly within the literature on coercive forces in dictatorships, particularly theoretical models that begin with the same two threats: ouster from groups outside the regime and coups from within.[19]

Much of the civil-military literature attempts to explain how various security institutions influence macro-outcomes such as regime instability, democratization, rebellions, and coups.[20] In these studies, the institutions governing civilian military relations are discussed, with the theories left to explain how different institutions shape regime outcomes. To date, however, there has been little effort to theorize how these institutions arise in the first place. Greitens’s book steps into this gap to examine how dominant threats to the regime influence the institutional structure of the security apparatus, and, in turn, how different institutional structures in the security forces influence violence. This is an important contribution because Greitens opens the black box that falls between threats and macro-outcomes such as regime stability. Before we can assess how threats to the regime influence regime stability, we first need to know how the main actors in the story, namely autocratic leaders, respond to threats by designing different types of security forces.


Greitens posits that leaders who perceive a high coup threat fragment their security forces and make them more exclusive. Fragmented and exclusive security forces protect the dictator from coups but also lead to state-led violence that is higher intensity, broader in scope, and less discriminate. However, when leaders believe the dominate threat to their rule stems from mass uprising, they unify the security forces and make them more inclusive. A unified and inclusive coercive apparatus yields less intense, narrower, and discriminate state-led repression.

A novel innovation in this account moves the motivating action from objective indicators of threat to the subjective threat perception of individual leaders. If we theorize from objective indicators—information that researchers glean from a bird’s eye view of dictatorships—we lose a relevant actor in the story, namely the leader who makes decisions about the institutional design of the coercive apparatus. The move to the leader’s threat perception is Pareto optimal from a theoretical standpoint because it is the more proximate causal location than objective threats; theorizing from threat perception is consequential when subjective perceptions diverge from objective indicators. The drawback, as Greitens points out, is the difficulty in accessing information about subjective threat perceptions.

Fragmentation and exclusivity increase violence through two mechanisms: intelligence and incentives. Fragmentation blocks the centralized flow of information necessary to accurately assess threats from outside the regime and exclusivity impedes information collection about out-groups. More accurate information, in turn, allows regimes to target opponents with selective, low-intensity repression. Fragmentation also creates organizational incentives that impede intelligence-gathering: competing organizations have an incentive to serve up inaccurate information; they report on each other instead of groups outside of the regime; and they create “obstacles to the vertical transmission of information” (50). Further, organizational competition creates direct incentives to increase indiscriminate violence to curry favor with the leader. With predictable consequences, exclusivity in the security forces lowers social sanctions for employing violent coercive power against out-groups. Each of these mechanisms points in the same direction: fragmentation and exclusivity increase violence.

A decision-theoretic model

While Greitens’s theory of institutional origins is interesting, it departs from existing literature by assuming there is no strategic interaction between the dictator and the security forces.[21] Instead, the decision-theoretic logic posits that leaders perceive one type of threat as higher than the other threat, and adjust their security forces optimally to counter the dominant one. Not only are security institutions sufficiently pliable in this account, but manipulating the institutional structure of these forces is costless to dictators. Leaders simply shift institutional designs because the security forces have no means to impose costs on the dictator.

A decision-theoretic model such as that of Grietens yields the prediction that leaders should invest in coup-proofing when coup risk is high. A strategic logic, however, suggests the opposite. Jun Koga Sudduth, for example, shows that leaders are more likely to implement coup-proofing changes to the security forces when perceived coup risk is low.[22] When dictators perceive a high coup threat, they have a disincentive to antagonize security elites by forcing changes on security institutions. Koga’s strategic logic yields this prediction because it allows security elites to bite back when leaders tread upon their interests, imposing a cost on leaders—in the form of staging a coup—for changing the security institutions.

The difference between a decision-theoretic and a strategic approach becomes apparent when assessing whether threat perceptions are endogenous to the strategies dictators employ to manage their security forces. Greitens posits that threat perception is generally exogenous, and can thus be employed as a proximate causal variable to explain institutional design.

Evidence from at least one case, however, suggests the opposite, and indeed is consistent with the strategic logic. A primary piece of evidence indicating Marcos’s perceived high coup threat is his diary entry dated from April 1972 (125). However, evidence of fragmentation comes from before 1972,[23] indicating that threat perception may have followed efforts to re-organize the security apparatus. Not only does the temporal sequence suggest that threat perception is endogenous in this case, but if threat perception was relatively low during the late 1960s into the early 1970s when Marcos’ began coup-proofing, then the timing of events is consistent with the strategic logic: coup-proofing is most likely when the coup risk is low and may increase grievance – and hence coup risk – in the short-run as security elites see their positions change.

When most researchers describe the security or guardianship dilemma faced by dictators, they imagine a dictator confronted with a trade-off (hence ‘dilemma’): moves to secure the regime against outside threats increase the capacity of the security forces to oust the leader; and moves to secure the leader against coups decrease the security forces’ capacity to put down or deter a rebellion.[24] A more accurate description of Greitens’s approach is “twin threats,” in which dictators assess the greatest threat (irrespective of the absolute level of threat) and adjust security forces accordingly. Future work can build on Greitens’s argument by incorporating her insights linking perceived threat to the incentives for institutional design of the security forces into a strategic theory that allows the security forces to retaliate when their interests are challenged.

Alternative explanations

Greitens juxtaposes her argument with several ‘big’ theories of institutional design: institutional inheritance and external influence. The former argues that security institutions are largely the path-dependent product of past iterations, while the latter posits that strong foreign actors, such as the U.S. in the aftermath of WWII, impose their own institutional design on weaker client states. The empirical evidence Greitens provides easily counters these explanations.

Greitens’s account, however, largely ignores the role of foreign actors in shaping threat perceptions in the first place. Theories of foreign influence encompass a wider framework than positive assimilation (states see the merits of a foreign model and copy it) or forced imposition (strong states force their model on a weaker ones). As important, foreign states shape the strategic environment. If a foreign power signals willingness to combat the main external threat faced by the regime – by intervening to counter mass protest (Saudi-Arabia/Bahrain, 2012), sending paratroopers to reverse a coup (France/Gabon, 1964), or deterring foreign invasion by stationing troops (U.S./South Korea post-1953) – the dictator perceives lower external threat and shapes his coercive forces accordingly. Indeed, when a foreign power substitutes for regime coercion in facing down an external threat, the dictator is better positioned to grab power from the security forces, easing the tension in the security dilemma. For example, foreign powers (the U.N. in the early 1960s, Belgium and France in the late 1970s) repeatedly saved the Mobutu regime in the former Zaire from armed rebellion, allowing the leader to design a poorly trained, fragmented, and exclusive security force. The threat of a coup was never far from Mobutu’s mind, but a coup never toppled him in over 30 years.[25] Instead, his regime disintegrated when foreign powers allowed a local crime boss to march youthful Ugandan soldiers across the country to Kinshasa, with Mobutu’s security forces fleeing in advance.

Greitens admits the influence of foreign powers when discussing the Korean military: the leaders lack the capacity to design military institutions to their liking because the U.S. directly controls this security institution. Thus, the relevant scope for testing Greitens’s argument in this case is the domestic intelligence and police forces. For a general theory of coercive institutional design to be compelling, it needs to either bracket the effect of foreign influence on strategic behavior by invoking ‘all else equal’ conditions or specify how this limits the scope conditions.

Case selection

The selection of cases nicely casts a wide net to include different types of autocracies: military juntas, personalist regimes, and dominant party dictatorships. The shadow cases (East Germany, Saddam Hussein’s regime in Iraq, and the Augusto Pinochet regime in Chile) broaden the regional scope. However, all the regimes under study, with the exception of Hussein’s, ended peacefully and were followed by relatively stable democracies. There is little discussion of how evidence culled from fallen regimes that end this way might influence the findings or provide scope conditions for the theory.

A fifth of the dictatorships that have existed since 1946 are still alive today, and among those that have collapsed, the majority have given way to a new dictatorial regime, not democracy. And while most transitions to democracy have sprung from largely non-violent protest movements and/or elections, the types of regimes discussed in this book comprise just over one-quarter of all dictatorial regimes since 1946. These descriptive statistics do not detract from the hard work of collecting archival data from collapsed regimes, but the design limits the scope. Indeed, in each of the four main cases and two of the three shadow cases, the security forces failed to defend the regime in the end. For this reason, the discussion of Hussein’s regime in Iraq is the most important and compelling. But even with this case, we have little purchase on whether the theory holds for dictatorships that end violently at the hands of domestic assailants, as Hussein’s regime was ousted by a foreign military super-power.

Comparative framework for evaluating evidence

Greitens provides an excellent conceptual discussion of fragmentation and exclusivity. Fragmentation, she contends, is the extent to which distinct security organizations have overlapping functions—on paper and in practice—and whether these organizations lack a “coordinating authority.” This definition moves beyond sociological accounts of civil-military relations (e.g. professionalism, civilian control) or institutional approaches that operationalize internal security politics as proxy in-group/out-group battles[26] or as numerical counts of organizations.[27]

Exclusivity is the extent to which “participation in the coercive apparatus is restricted” (27) to narrow groups defined by the country’s dominant social cleavage. This definition is broad enough to allow its application to ethnic, tribal, sectarian, religious and even regional or social class cleavages as they arise in different countries and contexts. For example, while ethnicity matters in Taiwan, it does not in South Korea, where regional differences matter.

High fragmentation often accompanies high exclusivity while low fragmentation (or unified forces) and low exclusivity (or inclusiveness) frequently co-occur. There are good theoretical reasons to expect this because these features re-inforce each other.

However, I was a bit lost in understanding how to interpret evidence that plausibly fits into different conceptual categories that are not expected to occur together, such as evidence that might be viewed as both fragmentation and inclusiveness.[28] For example, increasing the number of personnel in the security forces is evidence of exclusivity in the Philippines (133), but adding personnel to the domestic intelligence agency in South Korea under Park is interpreted as evidence of inclusivity (154). The existence of seven intelligence agencies in early in Park’s regime in Korea is evidence of inclusivity (154) but Marcos’s creation of multiple security organizations in 1971 is evidence of fragmentation in the Philippines (135). Later, in 1975, Marcos created a unitary joint command structure for the police (134), which Greitens interprets as evidence of fragmentation. In contrast, Park’s creation of the Defense Security Command in South Korea, which merged various armed service counter-intelligence units, is provided as evidence of fragmentation (163).

Before reviewing specific cases, the reader should know the general rules that led the researcher to interpret specific information, such additional personnel, as evidence of fragmentation rather than inclusion. More broadly, why are various types of information—the number of security personnel or organizations, the creation of joint command structures, or the appointment of officials from different organizations or military academy classes—interpreted differently in different contexts? Good social science that defines and measures concepts across different contexts requires a consistent standard for interpreting this information.


This book asks an important question: how do dictators design their security apparatus? To answer this question, Greitens posits that dictators shape the coercive apparatus in their states in response to the predominant threat faced by the regime. By moving the causal locus to the threat perceptions of autocratic leaders, Greitens’s book makes an important contribution to the literature. The theory linking perceived threats to institutional design strategies, however, is less compelling because the story assumes away the strategic interaction between the dictator and the security forces he manipulates. In short, the coercive apparatus has very little agency, and even less capacity to bite back when dictators tread on their interests. Future work can move the literature forward by incorporating the threat-perceptions approach pioneered in Greitens work into a strategic theory that takes seriously the tradeoffs presented by the security dilemma that is faced by all dictators.



Author’s Response by Sheena Chestnut Greitens, University of Missouri

It is a pleasure and a privilege to engage in a discussion of my book with the participants in this roundtable. I admire and have learned much from scholarship by each of the contributors, and am glad to have the opportunity to discuss their detailed and thoughtful reviews and suggestions. Below, I review the main findings of Dictators and Their Secret Police and summarize what I see as the book’s key potential contributions; I then discuss the questions raised by the reviewers and expand on the implications of this discussion, both for policy and for future research that could address some of the questions raised.

Dictators and Their Secret Police seeks to explain variations in repression that were difficult to account for using conventional theories, especially the “threat-response theory” or “law of coercive responsiveness” which suggest that increases in societal contention or opposition to authoritarian rule should be followed by corresponding increases in repression.[29] The book found that many of the patterns of state violence observed under authoritarian rule can be traced to variations in coercive institutional design, variations that in turn emerge from the threat perceptions of autocratic leaders at the time they assume power. Coup-proofing calls for fragmented and socially exclusive security organizations, while protecting against popular unrest demands unitary and inclusive ones. A fragmented, exclusive coercive apparatus gives its agents social and material incentives to escalate rather than dampen violence, and hampers agents from collecting the intelligence necessary to engage in targeted, discriminate, and pre-emptive repression; by contrast, a unitary and inclusive security apparatus configured to address significant mass unrest has much better intelligence capacity vis-à-vis its own citizens, and creates incentives for agents to minimize the use of violence and to rely instead on alternative forms of repression, such as surveillance and targeted pre-emption. These variations in coercive institutional design can alter or even invert the relationship between threat and repression; Dictators and Their Secret Police finds that regimes that are most threatened by popular contention will design institutions that end up employing violence that is comparatively lower-level, selective, and preventive.

One of my main objectives in writing Dictators and Their Secret Police was to more clearly conceptualize two kinds of variation that we observe in security behavior under authoritarian government: variation in forms of state repression, and variation in the internal security agencies, military and otherwise, that carry out that repression and violence on a day-to-day level. I am pleased to hear that the reviewers find the conceptualization of these variations to be strengths of the project, and that they see the theoretical framework used to explain them as making significant contributions to our understanding of authoritarianism, internal security, repression, and political violence. I also fully endorse their call to further investigate a number of variables, ideas, and questions that bear on the dynamics discussed in Dictators and Their Secret Police, but which the book for various reasons did not fully address and answer. Below I address the theoretical questions raised by the reviewers before turning to some empirical considerations.

Theoretical Questions

The reviewers raise several important theoretical questions about issues in the book, including its conceptualization of the choices that dictators face about coercive institutional design and other governing strategies; the agency of the security forces and whether a choice theoretic framework would be more appropriate; how to situate the military vis-à-vis the rest of the coercive or internal security apparatus; and the relationship between popular protest and military coup. Let me briefly address each of these in turn.

Coercive Institutions and Alternative Strategies

Several of the reviews raise different forms of the following question: are the choices facing dictators actually as stark as Dictators and Their Secret Police suggests? Jasen Castillo asks why a dictator cannot worry about both popular uprising and coup risks simultaneously, and suggests that my theory “defines coup proofing as a strategy that is incompatible with preventing popular uprising makes.” Risa Brooks raises a similar question: are the choices facing autocrats (between the fragmented and exclusive apparatus on one hand, and the unitary and inclusive one on the other) actually as stark a dichotomy as I suggest? Similarly, Van Jackson notes that the book does not fully incorporate coercive institutional design into the full menu of tools that dictators have available, such as co-optation.

The short answers to these valuable and important questions are as follows. The book’s presentation of an autocrat’s ‘coercive dilemma’ is simplified; in fact, we observe considerable complexity in the coup-proofing tactics of dictators that extends well beyond the book’s two main institutional variables (fragmentation and exclusivity). Moreover, the book’s predictions about coup-proofing leading to fragmentation and exclusivity are probabilistic, meaning that a high coup threat does not produce those outcomes (or those outcomes alone) 100% of the time. I note in the book (22, note 11) that the distinction between the two threats, and the coercive apparatus that each threat produces, has been stylized for analytical clarity, but that the framework “explicitly accounts for the fact that elite and popular threats commonly co-exist” (33). I also discuss briefly how autocrats may use patronage, co-optation, and other loyalty-inducing tactics for coup-proofing purposes (24). On this latter point, I use the example of Egypt, which Brooks agrees depended primarily on an economic bargain to maintain military loyalty under President Hosni Mubarak; additionally, some Communist regimes managed unitary and inclusive coercive institutions by employing high levels of political-party penetration in order to ensure loyalty and compliance and prevent coup threats.

In the book, I hypothesize that dictators might well worry about both popular and elite threats, but that there are specific characteristics of coercive institutional design where leaders would face a tradeoff. On these specific attributes, I posit that leaders would simply have to choose which threat they were most worried about; as a result, the “average” security apparatus of a dictator concerned principally with coup risk would have more fragmentation and higher exclusivity than that of a leader who was concerned more about popular overthrow. In order for Castillo’s and Brooks’s concerns to change these predictions, we would have to believe two things: first, that a dictator who was principally worried about coup risk would design a coercive apparatus that used economic payoffs and political penetration instead of fragmentation and exclusivity (rather than in addition to them), and second, that it is more common among dictatorships to rely exclusively on these alternative methods than it would be to incorporate the use of fragmentation and exclusivity. The evidence I have seen, both for the cases covered in the book, and for a wider set of cases that I examined, suggests that when these alternative tactics appear, they often accompany (rather than substitute for) fragmentation and exclusivity; and that they are the less frequently employed tactics than the two I focus on (perhaps because some research has shown that alternative tactics are less effective).[30] A full test to empirically validate what is admittedly an intuition on my part would be valuable, but requires cross-national data that does not yet exist (more on that point below).

A Difference in Dependent Variables: State Violence vs. Regime Stability

It is also important to be clear about when and how these alternative strategies would matter for the theoretical claims and outcome predictions made in the book. Dictators and Their Secret Police focuses particularly on fragmentation and exclusivity because these were aspects of coercive institutional design that I believed would have systematic downstream effects on state violence against civilians. The use and extent of alternative strategies, such as economic payoffs to security forces, may well be a critical causal variable shaping regime longevity or pattern of breakdown, but because the expected theoretical effect of those choices on violence against civilians was less clear, I chose not to examine the variation that undoubtedly exists—and that may be highly relevant to other outcomes. The point here is that the features of coercive institutional design that I find to be key explanatory factors in shaping patterns of state violence against civilians may well be very different from the factors that influence leader tenure, regime survival, or method of regime breakdown. I hope that future research takes up exactly these kinds of questions.

Moreover, to specifically address a question raised by Castillo, what I am saying also suggests that coup-proofing and popular management are not diametrically opposed strategies in general, but rather that they have tradeoffs in two very specific features of coercive institutional design that turn out to matter for violence against civilians. There is therefore a rich potential field of research that could examine the effect of various dimensions of coercive institutional design on a whole range of additional outcomes: the relationship between coercive institutional design and the presence of alternative strategies (institutions, economic patronage, and the like); how coercive institutional design affects regime or leader tenure; whether coercive institutional design is or is not correlated with the way a particular authoritarian regime fails; etc.

This overall line of inquiry, however, is important for the larger study of authoritarian politics and regime survival. If there are systematic exceptions to the link that I found between coup threat and fragmentation/exclusivity—if there exists a subcategory of cases where alternative strategies are systematically more common forms of coup-proofing than the two methods I examine in depth—then our theories should be revised and adjusted to account for this alternative path of institutional creation and its consequences. In other words, if there is a systematic element to the theory’s ‘off-the-line cases,’ where dominant perceived threat does not match coercive institutional design, it would be particularly valuable to identify that pattern for further theory generation and empirical testing.

Security Force Agency and Strategic Interaction

Risa Brooks and Joseph Wright both raise the issue of the agency of the security apparatus: whether it is as constrained by the decisions of the key leader as Dictators and Their Secret Police suggests. I agree that this is an important and valid question about the construction (and subsequent management) of the coercive apparatus, and appreciate the chance to clarify and expand on the temporal and, indeed, strategic logic that I see at work in dictators’ choices. I also agree that the book raises interesting questions that are not fully answered, and look forward to reading future scholarship that explores many of these questions.

Dictators and Their Secret Police suggests that dictators engage in coup-proofing via coercive institutional design during a short window after they first take power, because “if an autocrat is strong enough to seize power from the previous system [or leader], he is often strong enough to engage in significant institutional revision” (38). After this, however, I theorize that the coercive apparatus will become sticky and get locked in precisely because the agencies or their leaders are powerful political actors who have “both the capability and the desire to defend the institutional status quo to preserve [their] prerogatives” (63). I contend not that institutional manipulation is costless to autocrats, as Wright suggests, but that the costs vary over time. They are lower at the beginning of a leader’s tenure than they are later on in his rule; for that reason, a leader who anticipate the dominant threat to be a coup will attempt, in this early period of relative strength, to lock in institutional designs that will help them keep that threat in check. Based on what is or is not created in that initial period, the book also outlines the informational dynamics that should lead to institutional change over time; I theorize that only a strong signal to the autocrat that he has prioritized the wrong threat to his survival will lead him to attempt significant revisions to the coercive apparatus, and that a unitary and inclusive apparatus will be more likely to send that signal than a fragmented and exclusive one.

Thus while the logic in the book may read as decision-theoretic, I believe that there is more consistency with strategic explanations than it might seem upon initial reading. There is also a fair amount of empirical consistency in the predictions generated. Wright suggests, for example, that a strategic logic would lead to different empirical predictions about coup-proofing than the logic outlined in Dictators; I do not necessarily see this as the case. As an example, he highlights the work of Jun Koga Sudduth, whose writing was not available at the time I finished the manuscript, but with whom I find much to agree about the logic of dictatorship and coup-proofing.[31] Sudduth’s central empirical finding is that dictators are likely to purge in the first 3 years of their rule, which parallels two claims in my book: 1) that dictators engage in coup-proofing changes to the coercive apparatus early in their tenure, because that is when they are in a position of strength to do so (38); and 2) that because authoritarian takeovers often involve the forcible seizure of power, most of them begin with a period of purges, arrests, etc. Indeed, the temporal window that I identify for this phenomenon in Dictators and Their Secret Police—the first 3 years (61-62)—is identical to the one in Sudduth’s work published a year later.

Wright also raises the example of President Ferdinand Marcos in the Philippines, suggesting that the fact that threat perceptions recur after institutional design choices are made creates a potential endogeneity problem. I concur with Wright that threat perceptions become endogenous, but the book theorizes when and how this exact dynamic will occur: a leader who begins coup-proofing via fragmentation and exclusivity will set up an apparatus that only relays indicators of coup threat, rather than accurate information on popular threat; as a result, the initial choice to coup-proof tends to be self-reinforcing in leaders’ minds over time (61-63). I believe that the Marcos case actually demonstrates the utility of the bounded theory of institutional change proposed in Dictators, one that is relatively compatible with strategic approaches.[32]

Coercive Institutions and the Military

Brooks raises an additional consideration about my choice to broaden the aperture for analysis beyond the military (and therefore beyond civil-military relations) to study what I call “coercive institutions: the cluster of organizations collectively responsible for domestic intelligence and internal security” (21). I chose to adopt this more functionalist definition because I was interested in how tasks of domestic repression are allocated across different institutional frameworks, and therefore needed to test, rather than assume, whether distinctions between the military and non-military coercive actors mattered for the outcomes I was interested in. Brooks is entirely correct that there are often important sociological, organizational, and other differences between military and non-military actors. There are also authoritarian regimes where the distinctions between civilian security agencies and the military are clear, and the differences between the two are easily delineated—but part of my aim in this project was to show that scholars should not take that distinction for granted. The key questions are a) whether and to what extent there are systematic differences in the role of the military in repression, and b) how much any of those differences matters for explaining patterns of repressive behavior. For that reason, Dictators started with the premise that the role of the military in domestic coercive behavior should be theorized and tested, rather than assumed.

In examining who actually did the day-to-day work of repression in different authoritarian regimes, particularly those without a clear external threat where the military had the option of taking on a domestic repressive role, I found a surprising amount of organizational variation—variation in which the specific distinction between military and non-military organizations mattered less than I had initially expected. Repression in the Philippines was done by a combination of the police, military, and constabulary (a force that was at times a service branch of the Armed Forces of the Philippines (AFP) and at other times a domestic policing agency). In South Korea under Chun Doo-hwan, military conscripts ended up serving, in a near-historical accident, in the riot police, while the Korean Central Intelligence Agency (KCIA) handled both domestic and foreign intelligence work. In Taiwan, which was under martial law for decades, Taiwan Garrison Command (a military body) implemented internal security directives given by the national security organs of the president, but coordinated closely with police, intelligence, and other party organs at both the national and the local levels.

I sought therefore to make a serious attempt to characterize and theorize the organizational heterogeneity of the coercive apparatus under dictatorship, which meant placing the military in the overall context of repressive work. A fruitful direction for future research would be to ask precisely under what conditions the military, as an actor, does and does not engage in internal repressive work, and of what type, and why. Based on the theoretical logic outlined in the book, I would expect that the military (as opposed to civilian security agencies) is more likely to get involved with or take the lead in repression when two conditions are met: 1) external threat is relatively low, and 2) there is a need for parallel security forces based on the dictator’s perceived threat of a coup. In that case, competition between the police and the military is one option for fragmentation that serves coup-proofing purposes.[33] It is, of course, not the only form that fragmentation or counter-balancing can take: a dictator could also personalize either of these services, or create a separate (non-military, non-police) presidential security detachment, both of which are options that raise the likelihood of both fragmentation and exclusivity. Future research, then, could look at exactly what kind of conditions would prompt the military, as opposed to other types of coercive organizations, to engage in domestic repression, and under what conditions it does so as one of several fragmented organizations versus having a relative monopoly on repressive work. Future research could also test the implications of specific variants of fragmentation and patterns of military involvement in repression on outcomes including loyalty under crisis; regime durability; and performance vis-à-vis external threats.

Protest, Coup, and Regime Survival

Castillo raises an additional important point, which is the relationship between different potential pathways to regime breakdown, especially the relationship between popular protest and military coup (or at least security force defection). I noted above that regime breakdown is not a dependent variable that I attempted to explain in Dictators, but I did come away from the project with much greater appreciation of the need in comparative politics for more precise theorizing about the interaction of factors that create pathways to regime breakdown—as well as a need for more disaggregated coding of cross-national data on these processes. Datasets commonly code regime breakdown (democratic and autocratic alike) as either coup or demonstration/popular protest, but many of the cases that I have analyzed highlight the inter-relationship between these two phenomena. This is especially true for what Aries Arugay calls “civil society coups,” in which activism and protest by social forces prompt military intervention to restore stability (and in some cases, to address protestors’ grievances).[34]

Similarly, military or security force decisions are often a critical intervening variable that determines whether protest in the streets translates into autocratic breakdown. In South Korea, for example, the military’s reluctance to repress protests in 1987 helped convince Chun and his successor Roh Tae-woo to announce elections, while the 1986 ‘People’s Power Movement’ that toppled Marcos in February of 1986 is more accurately described as a combination of protest and a failed coup attempt led by two senior figures in the Philippine military (Fidel Ramos and Juan Ponce Enrile). Similarly, during the Arab Spring, security force behavior was a critical factor in determining the fate of protest movements: it was the military’s abandonment of President Zine El Abidine Ben Ali, for example, that paved the way for democratization in Tunisia, and the choices of the Egyptian military that led both to Mubarak’s fall and the current presidency of former military officer Abdel Fattah el-Sisi.[35] Yet in the Geddes, Wright, and Franz dataset, for example, these three regimes are all coded simply as ending by popular demonstration.[36] There is no mention of the inter-relationship between popular protest and security force behavior, or of the variation in that behavior that arguably shaped the subsequent transition: reluctance to act on the dictator’s behalf in the case of South Korea and Tunisia, versus an active attempt to oust or overthrow the incumbent in the Philippines and Egypt. Scholars of both security organizations and comparative authoritarianism could benefit from detailed and rigorous theorizing about the various pathways and interrelationships among coercive actors that lead to regime breakdown.

Additional Areas for Future Inquiry

Let me end this section with two other brief points where I am in full agreement with the reviewers. First, I wholeheartedly endorse Jasen Castillo’s call for a better understanding of how ethnic divisions influence both structure and staffing in the coercive apparatus. My key variable of interest was exclusivity, which could take on an ethnic form but did not have to (as in Korea, which is ethnically homogeneous, but where at least one of the two autocrats I analyzed used regional distinctions to make internal security personnel decisions). There is no doubt that this question is worth more systematic and detailed examination, including the manner in which different underlying levels of ethnic division in a society affect the optimality of different coercive institutional design choices. I am glad that in the time since the book has come out, some excellent scholarship has already begun to take up this task, and look forward to continued dialogue on these questions.[37]

Second, Van Jackson asks whether patterns of internal violence can become endogenous to the historical use of violence: in other words, whether the reputation of a coercive apparatus that is perceived by the population to be effective will have a deterrent effect on subsequent mobilization. I believe that the answer is yes: repressive competence does produce a deterrent effect, alongside an intelligence-based pre-emption mechanism. In other words, coercive institutions that consistently apprehend dissidents early and with accuracy (pre-emption) will eventually dissuade them from organizing and mobilizing at all (deterrence). I see this mechanism as fully compatible with the overall logic of the theory, and appreciate Jackson highlighting it in his review.

Empirical Concerns

Joseph Wright raises several questions about the empirics of the cases examined in Dictators and Their Secret Police, and whether my interpretation of the historical record raises questions for either the coding of variables or the case’s overall support for the book’s theoretical arguments. Let me note first that I agree entirely with his premise—that “good social science that defines and measures concepts across different contexts requires a consistent standard for interpreting this information” and then address each question in turn, as concisely as possible.

First, Wright suggests that the general rules that operationalize fragmentation and exclusion are inconsistent across cases. While historical reality does not always line up neatly with coding rules, the book’s theory chapter does clearly lay out the criteria for fragmentation: whether multiple organizations exist within the coercive apparatus without one of them having coordinating authority (25). If a single organization directs others and draws information from them upon demand, then a system with multiple organizations is coded as unitary rather than fragmented. This is why, for example, the creation of a joint command structure for the police did not resolve fragmentation in the Philippine coercive apparatus (134), and the merging of multiple military units into the Defense Security Command (DSC) (163) did not make South Korea’s coercive apparatus unitary: coercive power was still fragmented across the military, the police, and the Presidential Security Command (PSC) in the Philippines, and between the DSC and the KCIA in South Korea. These questions highlight the point raised above: it is critical to move beyond examining only the military, or only civilian agencies such as the police, because doing so can lead to inaccurate assessments of the coercive apparatus as a whole. Instead, scholars must examine the relationships among these different types of coercive actors.

Similarly, Dictators and Their Secret Police proposes that inclusivity is not simply a matter of numerical participation in the coercive apparatus (though that is one major indicator), but notes that these quantitative metrics can either be augmented or undercut by personnel practices such as deployment and funding (27-30). Marcos did initially increase the size of the military and police per capita, but the overall trend for the coercive apparatus as a whole during his tenure was negative (from 1:492 police per capita in the mid-1970s to 1:1120 by the 1980s). Chapter 4 further describes how deployment policies such as rotation and assignment of (one-third of) the armed forces to presidential security rather than counter-insurgency or popular policing means that they were not engaged in the task of dealing with popular threat (133-134). In the case of the Philippines, numbers, ethnic dynamics, and deployment patterns all combined to make the coercive apparatus relatively exclusive. Similarly, Wright mentions an example in Ch. 5 (154) that notes that seven different intelligence agencies reported on a single university campus in South Korea. Because this occurred in a context in which the KCIA had explicit coordinating authority over each of these organizations, and personnel embedded in many of them to ensure that authority (150), this example does not provide evidence of fragmentation; rather, it is evidence that the inclusivity of the domestic intelligence-gathering apparatus resulted in a fairly intensive, thorough penetration of Korean society.

Wright further raises the question of whether the peaceful ending of these regimes—and the fact that most of them resulted in democracy, whereas autocratic-to-autocratic transitions are a sizeable fraction of transition outcomes since 1946—poses problems for the scope of the argument. (The book’s four main cases include one violent transition, from Park to Chun after Park’s assassination in 1979-1980; one hereditary succession leadership change, from Chiang Kai-shek to Chiang Ching-kuo in Taiwan; and one abortive/failed coup attempt that combined with popular mobilization to end the Marcos regime in the Philippines in 1986.) In the book’s discussion of its empirical strategy (esp. 67-71), I noted that this combination of cases is the only one I could find where sufficient documentary evidence exists on all three variables of interest—the dictator’s subjective threat perceptions, coercive institutions, and patterns of state violence—to be able to answer the book’s central research question in a theoretically justifiable way. Cases where a country has democratized are far likelier to have pursued truth commissions, opened archives, and pursued transitional justice processes that make interviewees willing to speak candidly about the authoritarian period (70-71) so that the workings of the coercive apparatus can be accurately characterized—and equally important, so that accurate data on the nature and scale of state violence can be collected and analyzed. In my experience, cases where an authoritarian regime is ongoing seldom have accurate data on state violence, which was the core dependent variable of interest in this project.[38] I believe that the advantage gained by using historical cases to substantially increase data availability and data accuracy outweighs potential concerns about the post-transition trajectories of these systems.

I make this assessment in part because it is unclear to me exactly why the nature or outcome of these transitions would necessarily create problems for the book’s causal arguments about outcomes that occurred before that. This may be, in part, because my analytical interpretation of security force behavior in the endings of the regimes analyzed in Dictators is, as described in brief above, somewhat different from Wright’s. He characterizes all of the main cases as ones in which “the security forces failed to defend the regime.” This seems to me to brush away important analytical distinctions among the four cases, in terms of the behavior of the coercive apparatus and their relationship to regime demise. Park was assassinated by a member of his own intelligence service; Chun’s coercive apparatus in South Korea was reluctant to engage in repression, helping to trigger his successor’s decision on democratization; Taiwan’s coercive apparatus remained both loyal and capable but leaders acceded to popular demand for democratization for reasons that had little to do with coercive failure;[39] and Marcos left power after a split within the coercive apparatus triggered an abortive coup d’état attempt, a renewed wave of popular mobilization, and a loss of support from Marcos’ allies in the United States. This is fairly heterogeneous behavior by the security forces, which simply reinforces a point raised above—that the link between coercive institutions and regime outcomes is not one theorized by this book, but one which the field should more systematically address. Although it is always possible that the case selection in Dictators creates problems for the scope of the argument, it remains unclear exactly what kind of truncation or bias in the pre-transition data on coercive behavior occurs by choosing cases that have all transitioned to democracy, and done so (relatively) peacefully.

Finally, Wright suggests that Dictators and Their Secret Police “largely ignores the role of foreign actors in shaping threat perceptions in the first place.” I disagree; the book states that in fact, “external factors had the greatest influence indirectly, by influencing autocratic perceptions of the dominant threat” though they also had an especially direct role in shaping the South Korean military as an institution (295). I agree with his general point that it is important to be precise about the pathways by which foreign actors can and do influence the coercive apparatus and repressive behavior.[40]

Implications for Policy and Future Research

To my mind, this set of reviews raises two major questions—one for policy and one for future research—that are worth remarking on before concluding.

Van Jackson mentions the applicability of the book’s framework to current policy debates on American and international strategy toward North Korea. His comment that I appear to “downplay the explanatory potential of [my] argument for North Korea” is partly true: we lack reliable and comparable data on two of the three major variables of interest needed to make the argument made in the main sections of the book. Unlike the cases that Dictators and Their Secret Police focuses on, we lack information on the subjective threat perceptions of the Kim leaders, and we do not have anything close to reliable data on the scope and intensity of state violence against the population over time. I therefore chose to err on the side of caution when it came to what the book could and could not explain about North Korea.

I do think, however, that the framework advanced in the first half of the book helps to shed light on the construction and evolution of North Korea’s coercive apparatus under each of its three leaders, an argument that is mentioned in brief in the book (302) and that I hope to publish in full in the near future.[41] After eliminating elite threats relatively early in his tenure (by around the mid-1950s), Kim Il-sung focused on establishing intensive popular control; Jackson and I agree that Kim Jong-il, by contrast, emphasized elite threats and coup-proofing, which I assess as a function of his ascension to power via hereditary succession. The nature of Kim Jong-un’s coercive apparatus is still emerging; as I note in Dictators, the first years of any authoritarian regime are often chaotic, full of reorganization, personnel shuffling, and purges. My own assessment, however, parallels that of Jackson: that Kim Jong-un appears to have decided to follow more in Kim Il-sung’s footprints than Kim Jong-il’s. Violently eliminating elites with influence among the security forces and potential backing from China, for example, that Kim Jong-un seems to have copied straight from the Kim Il-sung playbook. That strategy also bears some resemblance to the course of events in Taiwan, where early elimination of elite rivals (including those viewed favorably by the United States) paved the way for Chiang Kai-shek to focus on popular control.

The question, then, is whether the changes that have been made to coercive institutional design in recent years in North Korea make Kim Jong-un more or less vulnerable to a coup.[42] Just because a leader attempts to coup-proof does not mean that those efforts will succeed. Sudduth suggests that purges can backfire if they incentivize remaining elites to coordinate to remove the leader, though early purges that successfully remove elite threats (e.g. Taiwan and Kim Il-sung’s North Korea) appear to contribute to regime longevity. Erica De Bruin’s work suggests that counter-balancing, which is one of the principal tools that Kim Jong-il employed, is one of the less effective techniques of coup-proofing,[43] but the extent to which Kim Jong Un will maintain or reverse this one of his father’s habits is as-yet unclear.

This is all to say that I strongly agree with Jackson that the book’s framework can be utilized to think about why Kim Jong-un is doing what he is doing and what the possible effects of certain choices on regime stability might be. At the very least, it points toward some new and different indicators that could help shed light on these questions and inform forecasting by both the U.S. government and the broader international community. I am surprised to learn from Jackson that coup risk has not been a priority area of analysis among policymakers, and hope that my own work and that of the scholars working on these questions will soon mean that these variables are incorporated into scenarios and other policy planning exercises related to policy on and strategy toward North Korea.

The last point that this discussion has highlighted is the need for large-scale data collection on the coercive institutions of dictatorships, both historically and in the present moment. Questions about regime survival would be best answered by testing the effect of various more specific aspects of coercive institutional design, including those examined in Dictators, on outcomes such as regime duration, leader tenure, and the manner in which a leader exits power or a regime breaks down. The data on these dependent variables already exists, but large-scale, rigorously coded cross-national data on the coercive apparatus of authoritarian regimes (and potentially someday of democracies) would be a useful contribution to multiple fields of political science, as well as to contemporary policy questions on security, human rights, and American foreign policy. I am currently working on proposals to begin this data collection, and hope to combine forces with similarly-interested researchers to construct this data for the field’s long-term use.

Lastly, let me again thank the participants in this roundtable for their insightful and constructive reviews of Dictators and Their Secret Police. Participating in this conversation has advanced my own thinking about these issues and questions, and I look forward to seeing where the discussion and research goes next.



[1] Some of the literature on the Arab Spring has also tried to address these trade-offs. For example, see Hicham Bou Nassif, “Generals and Autocrats: How Coup-Proofing Predetermined the Military Elite’s Behavior in the Arab Spring,” Political Science Quarterly 130:2 (Summer 2015): 245–275. DOI:

[2] Risa Brooks, “Military Defection and the Arab Spring,” Oxford Research Encyclopedia of Politics (February 2017): 19-20. DOI:

[3] Joshua Stacher’s terrific book gets at some of these dynamics; Stacher, Adaptable Autocrats: Regime Power in Egypt and Syria (Stanford: Stanford University Press, 2012). Also see Yezid Sayigh, Above the State: The Officer’s Republic in Egypt (Washington, D.C.: Carnegie Endowment for International Peace, 2012); Brooks, “Military Defection and the Arab Spring,” 9.

[4] Greitens addresses some potential endogeneities, such as that coercive institutions might be the result of regime type, or how a leader came to power (36). She addresses path-dependence and debunks that prior institutions are necessarily predictive of future institutions, especially when underlying conditions change. She also allows for endogeneity in perceived coup threat to fragmented institutions, arguing that this explains why autocrats may continue to perceive coups threats when they have this type of institutional set-up (63). She does not otherwise appear to address that the absence of perceived coup threat might be affected by the nature of coercive institutions.

[5] They also engage in external defense as Greitens acknowledges (30). On the relationship between coup-proofing mechanisms and militaries’ external roles in war-fighting see Caitlin Talmadge, The Dictator’s Army: Battlefield Effectiveness in Authoritarian Regimes (Ithaca: Cornell University Press, 2015).

[6] In fact, the issue of military-police relations is arguably under-studied.

[7] Risa Brooks, “Abandoned at the Palace: Why the Tunisian Military Defected from the Ben Ali Regime in January 2011,” Journal of Strategic Studies 36:2 (February 2013): 205-220. DOI: For a recent overview of methods of coup-proofing see Erica De Bruin, “Coup-Proofing for Dummies,” Foreign Affairs, 27 July 2014,

[8] Zoltan Barany, “Comparing the Arab Revolts: The Role of the Military,” Journal of Democracy 22:4 (October 2011): 24-35; Jason Brownlee, Tarek Masoud, and Andrew Reynolds, The Arab Spring: Pathways of Repression and Reform (Oxford: Oxford University Press, 2015); On cases in Asia see Terrence Lee, Defect or Defend: Military Responses to Popular Protests (Baltimore: Johns Hopkins Press, 2015).

[9] See Stepan’s seminal work. Alfred Stepan, Re-Thinking Military Politics (Princeton: Princeton University Press, 1988); Risa Brooks, Shaping Strategy: The Civil-Military Politics of Strategic Assessment (Princeton: Princeton University Press, 2008).

[10] For the effects of coup proofing on combat performance see Caitlin Talmadge, The Dictator’s Army: Battlefield Effectiveness in Authoritarian Regimes (Ithaca: Cornell University Press, 2015).

[11] See Mohammed Ayoob, The Third World Security Predicament: State Making, Regional Conflict, and the International System (Boulder: Lynne Rienner, 1995); and Stanislav Andreski, Wars, Revolutions, Dictatorships: Studies of Historical and Contemporary Problems from a Comparative Viewpoint (London, Frank Cass, 1992), especially 177-178.

[12] David M. Glantz and Jonathan House, When Titans Clashed: How the Red Army Stopped Hitler (Lawrence: University Press of Kansas, 1995).

[13] On Nazi Germany see MacGregor Knox, Common Destiny: Dictatorship, Foreign Policy, and War in Fascist Italy and Nazi Germany (Cambridge: Cambridge University Press, 2000). On the Soviet Union see Richard Overy, Russia’s War: A History of the Soviet War Effort, 1941-1945 (London: Penguin Books, 1997). A good overview of both regimes is Timothy Snyder, Bloodlands: Europe Between Hitler and Stalin (New York: Basic Books, 2012).

[14] See the discussion in Jasen J. Castillo, Endurance and War: The National Sources of Military Cohesion (Stanford: Stanford University Press, 2014).

[15] Andrew Salmon, “North Korean intelligence official tells of aborted coups and assassination attempts,” The Telegraph, 2 April 2014,

[16] K.J. Kwon and Ben Westcott, “Kim Jong Un has executed over 300 people since coming to power,” CNN, 29 December 2016,

[17] Tom Malinowski, “How to Take Down Kim Jong Un,” Politico, 24 July 2017,

[18] Doyun Kim, “Technology meets human rights at N. Korea hackathon,” NK News, 15 August 2014,

[19] Variants of the model set-up treat threat from outside as mass citizen uprising, armed rebellion, or a foreign state’s military. See Daron Acemoglu, Davide Ticchi, and Andrea Vindigni, “A theory of military dictatorships,” American Economic Journal: Macroeconomics 2:1 (2010): 1-42; Milan W. Svolik, The Politics of Authoritarian Rule (New York: Cambridge University Press, 2012); R. Blake McMahon and Branislav L. Slantchev. “The Guardianship Dilemma: Regime Security through and from the Armed Forces,” American Political Science Review 109:2 (2015): 297-313; Jun Koga Sudduth, “Strategic Logic of Elite Purges in Dictatorships,” Comparative Political Studies 50:13 (2017): 1768-1801; Jun Koga Sudduth, “Coup Risk, Coup-proofing and Leader Survival,” Journal of Peace Research 54:1 (2017): 3-15.

[20] Samuel E. Finer, The Man on Horseback: The Role of the Military in Politics (London: Pall Mall Press, 1962); James T. Quinlivan, “Coup-proofing: Its Practice and Consequences in the Middle East,” International Security 24:2 (1999): 131-165; Aaron Belkin and Evan Schofer. “Toward A Structural Understanding of Coup Risk,” Journal of Conflict Resolution 47:5 (2003): 594-620; Eva Bellin, “The Robustness of Authoritarianism in the Middle East: Exceptionalism in Comparative Perspective,” Comparative Politics (2004): 139-157; Terrence Lee, “The Armed Forces and Transitions from Authoritarian Rule: Explaining the role of the military in 1986 Philippines and 1998 Indonesia,” Comparative Political Studies 42:5 (2009): 640-669; Zoltan D. Barany. The Soldier and the Changing State: Building Democratic Armies in Africa, Asia, Europe, and the Americas (Princeton: Princeton University Press, 2012); Jonathan Powell, “Determinants of the Attempting and Outcome of Coups d’état.” Journal of Conflict Resolution 56:6 (2012): 1017-1040; Erica De Bruin, “Preventing Coups d’état: How Counterbalancing Works,” Journal of Conflict Resolution (2017): 1433-1458.

[21] Some of the earlier literature on domestic protest and repression also employs decision-theoretic logic. Mark Irving Lichbach, “Deterrence or Escalation? The Puzzle of Aggregate Studies of Repression and Dissent,” Journal of Conflict Resolution 31:2 (1987): 266-297; Christian Davenport, “Multi-Dimensional Threat Perception and State Repression: An Inquiry into Why States Apply Negative Sanctions,” American Journal of Political Science 39:3 (1995): 683-713.

[22] Sudduth, “Strategic Logic of Elite Purges in Dictatorships’; Sudduth, “Coup Risk, Coup-Proofing and Leader Survival.”

[23] “Even before 1972, Marcos had moved to make the top levels of the security apparatus more exclusive,” (128) See also, the cartoon dated June 1970 (131).

[24] See, for example, the principal agent models in Acemoglu, Ticchi, and Vindigni; Svolik; and McMahon and Slantchev.

[25] Mobutu purged his most competent—and popular—military commander, General Mahele, in 1992.

[26] Samuel Decalo, Coups and Army Rule in Africa: Studies in Military Style. (New Haven: Yale University Press, 1976); Philip Roessler. Ethnic Politics and State Power in Africa: the Logic of the Coup-Civil War Trap. (New York: Cambridge University Press, 2016); Kristen A. Harkness, When Soldiers Rebel: Ethnic Armies and Political Instability in Africa. (New York: Cambridge University Press, 2018).

[27] Belkin and Schofer; Tobias Böhmelt and Ulrich Pilster, “The Impact of Institutional Coup-Proofing on Coup Attempts and Coup Outcomes,” International Interactions 41:1 (2015): 158-182; De Bruin.

[28] This reader would also like to see information on the sources for the author’s calculation of security personnel numbers in Table 1.2 and Tables 5.1 to 5.3. Footnotes and the Appendix provides no information on the sources used for these calculations.

[29] Christian Davenport, “State Repression and Political Order,” Annual Review of Political Science 10 (2007): 1-23; Jenifer Earl, Sarah Soule, and John McCarthy, “Protest Under Fire: Explaining the Policing of Protest,” American Sociological Review 68:4 (2003): 581-606.

[30] Theo McLauchlin, “Loyalty Strategies and Military Defection in Rebellion,” Comparative Politics 42:3 (2010): 333-350; Michael Makara, “Coup-Proofing, Military Defection, and the Arab Spring,” Democracy and Security 9:3 (2013): 334-359.

[31] Jun Koga Sudduth, “Strategic Logic of Elite Purges in Dictatorships,” Comparative Political Studies 50:13 (2017): 1768-1801.

[32] It also seems analytically problematic to compare, as Wright does, a supposedly low coup risk in the late 1960s, when Marcos was a democratically elected president, to the post-1972 period when the Philippines was an authoritarian regime under martial law. The coup-proofing tools available to Marcos before the declaration of martial law were quite different than those available afterward, as the book details in Chapters 4 and 7.

[33] On the efficacy (or not) of this strategy, see Erica De Bruin, “Preventing Coups d’Etat: How Counter-balancing Works,” Journal of Conflict Resolution 62:7 (2018): 1433-1458.

[34] Aries Arugay, “Democracy’s Saviors and Spoilers: A Study of the Causal Conditions and Mechanisms behind “Civil Society Coups” after the Third Wave of Democratization,” PhD dissertation, Georgia State University (2014).

[35] Jason Brownlee, Tarek Masoud, and Andrew Reynolds, The Arab Spring: Pathways of Repression and Reform (New York: Oxford University Press, 2014).

[36] The Autocratic Regimes codebook for the Geddes, Wright, and Franz (2014) dataset is available at

[37] Among others, see McLauchlin, “ Loyalty Strategies and Military Defection in Rebellion”; Makara, “ Coup-Proofing, Military Defection, and the Arab Spring”; Mai Hassan, “The Strategic Shuffle: Ethnic Geography, the Internal Security Apparatus, and Elections in Kenya,” American Journal of Political Science 61:2 (2017): 382-395; Philip Roessler, “The Enemy Within: Personal Rule, Coups, and Civil War in Africa,” World Politics 63:2 (2011): 300-346.

[38] See, for example, the discussion in Ch. 6 of the inaccuracy of contemporaneous reporting on state violence in Taiwan (180-184).

[39] This assessment is shared by Steve Levitsky and Lucan Way, Competitive Authoritarianism (New York: Cambridge University Press, 2010); Dan Slater and Joseph Wong, “The Strength to Concede: Ruling Parties and Democratization in Developmental Asia,” Perspectives on Politics 11:3 (2013):717-733.

[40] Jeremy Kuzmarov, Modernizing Repression: Police Training and Nation-Building in the American Century (University of Massachusetts Press, 2012); Jessica Trisko Darden, Aiding and Abetting: U.S. Foreign Assistance and State Violence (Stanford: Stanford University Press, 2019).

[41] Sheena Chestnut Greitens, “The Construction of North Korea’s Internal Security Apparatus,” working paper (2018).

[42] Here, precise theory and empirical evidence are lacking: since I did not attempt to use the book’s framework to predict regime or leader survival, my answer here is necessarily speculative.

[43] Jun Koga Sudduth, “Coup Risk, Coup-Proofing, and Leader Survival,” Journal of Peace Research 54:1 (2017): 3-15; Erica De Bruin, How to Prevent Coups d’état: Coercive Institutions and Regime Survival (book manuscript in progress).